Dual Board Review

[Guest]

I found the Dual board reivew a little Windows centric and that the conclusions do not apply for other OS.

Both Be OS and Liunx opperate very well with multi processor systems with a high efficency - my own measurements with Be Os show it to be about 85% efficient (based on comparing SETI units over many workunits) - the losses seem to be associated with threads switching beteewn CPUs (Be OS does not allow thread to CPU assignment AFAIK). The result is that I can crunch 2 SETIs, and WP/Spreadsheet/Image/gameplay,whatever without any apparent (there must be some as there are only so many CPU cycles to go around) slow responses for manual tasks.

I think that to get the best out multi CPU boards it is better to run an OS that has been designed for the task.

Cheers

Neil

 

[Malc]

I'm personally very happy with Windows and my dual CPU machine. I do a lot of stuff that is very CPU and even disk intensive. The machine always stays very responsive. I also do a lot of stuff via PcAnywhere. Those remote machines have single CPU at almost twice the speed of my individual CPUs... I start MSVC compiling, or long query in SQL Server on those boxes and they grind to a halt! I can't do a thing on them until the job has finished. The start menu can take 30 secs to appear. I never want to go back to a single CPU machine. The only drawback that I see with dual CPU machines is the price over head (as mentioned in the article)... I'm holding off upgrading my 2 P2-450s until P3-850s drop below USD$150. For now I keep things speedy with lots of RAM allowing a huge disk cache and so my system isn't bottled-necked by the IDE disk sub-system. Windows really isn't that bad for SMP *workstations*.

 

[Guest]

Dear Malc,

This is not a popularity contest for Windows against other OS.
I don't think Windows that bad. It just for this article it more be suitable to use
the OS that were really meant for SMP.

And the article doens't really prove anything.
It compare a retail board with beta board.
Doesn't have conclusion?

It will be more usefull for reader if THG compare it
with old board such as Asus P2B-D, Abit BP6, Abit VP6, i840 board, i820 board.

And because this not about comparing between board,
the next good things THG could bring is comparing between OS.
Linux kernel 2.3 vs 2.4, NT 4.0 vs. Win2000, Win vs. Linux vs. BeOS.
Show how good each OS doing an SMP things without an apps crafted for it.
That's more real world to me.

PS: I don't like the Linux Kernel compilation benchmark on THG.
I don't really see the point of using it.
I just saw it as THG want to get a sympathy from Linux user.

PPS: I already mailed THG around 3 times about comparing different version of Linux
Kernel for SMP comparison, NT 4.0, Win2000 & BeOS. But I feel my mail go to /dev/null.

Get BeOS (http://free.be.com)
The Best ever Operating System

 

[tfbww]

It most certainly does show something. His point is that SMP is not valuable for Windows users, regardless of the board type, new or old. THAT is a valuable piece of advice since most people are stuck using Windoze.

That being said, a second article showing possible performance benefits with other OSes would be fine but would be a separate point.

 

[Guest]

It most certainly does show something. His point is that SMP is not valuable for Windows users, regardless of the board type, new or old. THAT is a valuable piece of advice since most people are stuck using Windoze.

<b>That point is most certainly Wrong</b>. Dual boards are great for windows 2000. With those cheap SMP boards (if they are really ok -hard to tell from the short article with no stabilty tests, etc.) and all-time low CPU prices, for only $340 more now (~5% of the system cost) even if you don't use high end software like Max or Photoshop, you can have a huge increase in responsiveness and performance when running just two applications at the same time. Even when just running one app, the os load is put on the other CPU so you still have a gain.

 

[Guest]

>That point is most certainly Wrong. Dual boards are great for windows 2000.
>With those cheap SMP boards (if they are really ok -hard to tell from the short
>article with no stabilty tests, etc.) and all-time low CPU prices, for only $340
>more now (~5% of the system cost) even if you don't use high end software
>like Max or Photoshop, you can have a huge increase in responsiveness and
>performance when running just two applications at the same time. Even
>when just running one app, the os load is put on the other CPU so you still have a gain.

This is why I want THG to test the SMP board with non-SMP software.
Please tell readers more than just Bapco Sysmark. Tell readers how it's feel.
Especially on the slow/big Win2000 that most reader were used.
And tell readers if it's still the same felling when using NT 4.0 or Linux.
We want more than just benchmark. <-- This is why I read THG in the first place.

Get BeOS (http://free.be.com)
The Best ever Operating System

 

[Guest]

I agree with Cycnuskus that the review was defective. Not only would it have been more appropriate to compare the different kernels of Linux, and BeOS, etc, it was minimally deceptive to highlight a topic of keen interest (multiple cpu's), and then ignore any testing which would highlight the deficiencies with Windows. I especially disagree with the conclusion that the additional cost of the second cpu nullifies the minuscule performance gain. That conclusion would have been unwarranted, even if the discrepancy between single cpu and dual cpu had been highlighted better by employing a conventional, inexpensive graphics controller so that the additional cpu's cost could have been accurately compared to the performance gain associated with multiprocessing. Use of a beta product, as noted by Cycnuskus, was ill advised. I have one postitive note to add about the review. It was completed in such a careless fashion that I rushed out and joined this forum to express my dismay. I have read the reviews at this site for nearly two years, without feeling a similar desire to speak out. I do not know if there are any others who feel as disappointed by this review, but I at least am not discouraged for the long run--ergo my decision to join the fray and enter the forum. Regards.

 

[Guest]

Dear hdnb,

If you read all my post in THG forum. It's all about dissapointment with THG article. I have read THG about
22 months. And only recently I felt the article quality went down.

I hope THG will hear their reader opinion to make them the best.

Get BeOS (http://free.be.com)
The Best ever Operating System